potok/opencv
1
0
Fork 0
mirror of https://github.com/opencv/opencv.git synced 2026-07-31 08:13:04 +04:00
Code Packages Releases Wiki Activity

Merge remote-tracking branch 'upstream/3.4' into merge-3.4

Browse Source
This commit is contained in:
Alexander Alekhin
2020-02-20 19:45:21 +03:00
parent 1f695c4532 150c29356a
commit 96b26dc8f4
28 changed files with 1115 additions and 147 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/dnn/src/caffe/caffe_importer.cpp
+67 -31
View File
@@ -75,6 +75,17 @@ static cv::String toString(const T &v)
return ss.str();
}
static inline
MatShape parseBlobShape(const caffe::BlobShape& _input_shape)
{
MatShape shape;
for (int i = 0; i < _input_shape.dim_size(); i++)
{
shape.push_back((int)_input_shape.dim(i));
}
return shape;
}
class CaffeImporter
{
caffe::NetParameter net;
@@ -235,10 +246,7 @@ public:
}
else if (pbBlob.has_shape())
{
const caffe::BlobShape &_shape = pbBlob.shape();
for (int i = 0; i < _shape.dim_size(); i++)
shape.push_back((int)_shape.dim(i));
shape = parseBlobShape(pbBlob.shape());
}
else
shape.resize(1, 1); // Is a scalar.
@@ -334,12 +342,49 @@ public:
//setup input layer names
std::vector<String> netInputs(net.input_size());
std::vector<MatShape> inp_shapes;
{
for (int inNum = 0; inNum < net.input_size(); inNum++)
int net_input_size = net.input_size();
for (int inNum = 0; inNum < net_input_size; inNum++)
{
addedBlobs.push_back(BlobNote(net.input(inNum), 0, inNum));
netInputs[inNum] = net.input(inNum);
}
if (net.input_dim_size() > 0) // deprecated in Caffe proto
{
int net_input_dim_size = net.input_dim_size();
CV_Check(net_input_dim_size, net_input_dim_size % 4 == 0, "");
CV_CheckEQ(net_input_dim_size, net_input_size * 4, "");
for (int inp_id = 0; inp_id < net_input_size; inp_id++)
{
int dim = inp_id * 4;
MatShape shape(4);
shape[0] = net.input_dim(dim);
shape[1] = net.input_dim(dim+1);
shape[2] = net.input_dim(dim+2);
shape[3] = net.input_dim(dim+3);
inp_shapes.push_back(shape);
}
}
else if (net.input_shape_size() > 0) // deprecated in Caffe proto
{
int net_input_shape_size = net.input_shape_size();
CV_CheckEQ(net_input_shape_size, net_input_size, "");
for (int inp_id = 0; inp_id < net_input_shape_size; inp_id++)
{
MatShape shape = parseBlobShape(net.input_shape(inp_id));
inp_shapes.push_back(shape);
}
}
else
{
for (int inp_id = 0; inp_id < net_input_size; inp_id++)
{
MatShape shape; // empty
inp_shapes.push_back(shape);
}
}
}
for (int li = 0; li < layersSize; li++)
@@ -364,6 +409,17 @@ public:
addedBlobs.back().outNum = netInputs.size();
netInputs.push_back(addedBlobs.back().name);
}
if (layer.has_input_param())
{
const caffe::InputParameter &inputParameter = layer.input_param();
int input_shape_size = inputParameter.shape_size();
CV_CheckEQ(input_shape_size, layer.top_size(), "");
for (int inp_id = 0; inp_id < input_shape_size; inp_id++)
{
MatShape shape = parseBlobShape(inputParameter.shape(inp_id));
inp_shapes.push_back(shape);
}
}
continue;
}
else if (type == "BatchNorm")
@@ -424,35 +480,15 @@ public:
}
dstNet.setInputsNames(netInputs);
std::vector<MatShape> inp_shapes;
if (net.input_shape_size() > 0 || (layersSize > 0 && net.layer(0).has_input_param() &&
net.layer(0).input_param().shape_size() > 0)) {
int size = (net.input_shape_size() > 0) ? net.input_shape_size() :
net.layer(0).input_param().shape_size();
for (int inp_id = 0; inp_id < size; inp_id++)
if (inp_shapes.size() > 0)
{
CV_CheckEQ(inp_shapes.size(), netInputs.size(), "");
for (int inp_id = 0; inp_id < inp_shapes.size(); inp_id++)
{
const caffe::BlobShape &_input_shape = (net.input_shape_size() > 0) ?
net.input_shape(inp_id) :
net.layer(0).input_param().shape(inp_id);
MatShape shape;
for (int i = 0; i < _input_shape.dim_size(); i++) {
shape.push_back((int)_input_shape.dim(i));
}
inp_shapes.push_back(shape);
if (!inp_shapes[inp_id].empty())
dstNet.setInput(Mat(inp_shapes[inp_id], CV_32F), netInputs[inp_id]);
}
}
else if (net.input_dim_size() > 0) {
MatShape shape;
for (int dim = 0; dim < net.input_dim_size(); dim++) {
shape.push_back(net.input_dim(dim));
}
inp_shapes.push_back(shape);
}
for (int inp_id = 0; inp_id < inp_shapes.size(); inp_id++) {
dstNet.setInput(Mat(inp_shapes[inp_id], CV_32F), netInputs[inp_id]);
}
addedBlobs.clear();
}
modules/dnn/src/dnn.cpp
+25 -7
View File
@@ -1418,13 +1418,15 @@ struct Net::Impl
clear();
this->blobsToKeep = blobsToKeep_;
allocateLayers(blobsToKeep_);
MapIdToLayerData::iterator it = layers.find(0);
CV_Assert(it != layers.end());
it->second.skip = netInputLayer->skip;
initBackend();
initBackend(blobsToKeep_);
if (!netWasAllocated)
{
@@ -1437,7 +1439,6 @@ struct Net::Impl
}
netWasAllocated = true;
this->blobsToKeep = blobsToKeep_;
if (DNN_NETWORK_DUMP > 0)
{
@@ -1564,7 +1565,7 @@ struct Net::Impl
ldOut.consumers.push_back(LayerPin(inLayerId, outNum));
}
void initBackend()
void initBackend(const std::vector<LayerPin>& blobsToKeep_)
{
CV_TRACE_FUNCTION();
if (preferableBackend == DNN_BACKEND_OPENCV)
@@ -1574,7 +1575,7 @@ struct Net::Impl
else if (preferableBackend == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019)
{
#ifdef HAVE_INF_ENGINE
initInfEngineBackend();
initInfEngineBackend(blobsToKeep_);
#else
CV_Assert(false && "This OpenCV version is built without Inference Engine API support");
#endif
@@ -1582,7 +1583,7 @@ struct Net::Impl
else if (preferableBackend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH)
{
#ifdef HAVE_DNN_NGRAPH
initNgraphBackend();
initNgraphBackend(blobsToKeep_);
#else
CV_Error(Error::StsNotImplemented, "This OpenCV version is built without support of Inference Engine + nGraph");
#endif
@@ -1688,7 +1689,7 @@ struct Net::Impl
}
}
void initInfEngineBackend()
void initInfEngineBackend(const std::vector<LayerPin>& blobsToKeep_)
{
CV_TRACE_FUNCTION();
CV_Assert_N(preferableBackend == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019, haveInfEngine());
@@ -1878,6 +1879,15 @@ struct Net::Impl
CV_Assert(!ieNode.empty());
ieNode->net = net;
for (const auto& pin : blobsToKeep_)
{
if (pin.lid == ld.id)
{
ieNode->net->addOutput(ieNode->layer.getName());
break;
}
}
// Convert weights in FP16 for specific targets.
if ((preferableTarget == DNN_TARGET_OPENCL_FP16 ||
preferableTarget == DNN_TARGET_MYRIAD ||
@@ -1984,7 +1994,7 @@ struct Net::Impl
}
}
void initNgraphBackend()
void initNgraphBackend(const std::vector<LayerPin>& blobsToKeep_)
{
CV_TRACE_FUNCTION();
CV_Assert_N(preferableBackend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH, haveInfEngine());
@@ -2173,6 +2183,14 @@ struct Net::Impl
// TF EAST_text_detection
ieNode->net->setUnconnectedNodes(ieNode);
}
for (const auto& pin : blobsToKeep_)
{
if (pin.lid == ld.id)
{
ieNode->net->addOutput(ieNode->node->get_friendly_name());
break;
}
}
ieNode->net->setNodePtr(&ieNode->node);
net->addBlobs(ld.inputBlobsWrappers);
modules/dnn/src/ie_ngraph.cpp
+4 -5
View File
@@ -231,11 +231,10 @@ void InfEngineNgraphNet::init(Target targetId)
}
}
}
} else {
for (const auto& name : requestedOutputs)
{
cnn.addOutput(name);
}
}
for (const auto& name : requestedOutputs)
{
cnn.addOutput(name);
}
for (const auto& it : cnn.getInputsInfo())
modules/dnn/src/layers/crop_and_resize_layer.cpp
+44 -5
View File
@@ -5,6 +5,7 @@
// Copyright (C) 2018, Intel Corporation, all rights reserved.
// Third party copyrights are property of their respective owners.
#include "../precomp.hpp"
#include "../ie_ngraph.hpp"
#include "layers_common.hpp"
#ifdef HAVE_CUDA
@@ -25,6 +26,14 @@ public:
outHeight = params.get<float>("height");
}
virtual bool supportBackend(int backendId) CV_OVERRIDE
{
return backendId == DNN_BACKEND_OPENCV
|| backendId == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH
|| backendId == DNN_BACKEND_CUDA
;
}
bool getMemoryShapes(const std::vector<MatShape> &inputs,
const int requiredOutputs,
std::vector<MatShape> &outputs,
@@ -41,11 +50,6 @@ public:
return false;
}
virtual bool supportBackend(int backendId) CV_OVERRIDE
{
return backendId == DNN_BACKEND_OPENCV || backendId == DNN_BACKEND_CUDA;
}
void forward(InputArrayOfArrays inputs_arr, OutputArrayOfArrays outputs_arr, OutputArrayOfArrays internals_arr) CV_OVERRIDE
{
CV_TRACE_FUNCTION();
@@ -121,6 +125,41 @@ public:
}
}
#ifdef HAVE_DNN_NGRAPH
virtual Ptr<BackendNode> initNgraph(const std::vector<Ptr<BackendWrapper> >& inputs,
const std::vector<Ptr<BackendNode> >& nodes) CV_OVERRIDE
{
// Slice second input: from 1x1xNx7 to 1x1xNx5
auto input = nodes[0].dynamicCast<InfEngineNgraphNode>()->node;
auto rois = nodes[1].dynamicCast<InfEngineNgraphNode>()->node;
std::vector<size_t> dims = rois->get_shape(), offsets(4, 0);
offsets[3] = 2;
dims[3] = 7;
auto lower_bounds = std::make_shared<ngraph::op::Constant>(ngraph::element::i64,
ngraph::Shape{offsets.size()}, offsets.data());
auto upper_bounds = std::make_shared<ngraph::op::Constant>(ngraph::element::i64,
ngraph::Shape{dims.size()}, dims.data());
auto strides = std::make_shared<ngraph::op::Constant>(ngraph::element::i64,
ngraph::Shape{dims.size()}, std::vector<int64_t>((int64_t)dims.size(), 1));
auto slice = std::make_shared<ngraph::op::v1::StridedSlice>(rois,
lower_bounds, upper_bounds, strides, std::vector<int64_t>{}, std::vector<int64_t>{});
// Reshape rois from 4D to 2D
std::vector<size_t> shapeData = {dims[2], 5};
auto shape = std::make_shared<ngraph::op::Constant>(ngraph::element::i64, ngraph::Shape{2}, shapeData.data());
auto reshape = std::make_shared<ngraph::op::v1::Reshape>(slice, shape, true);
auto roiPooling =
std::make_shared<ngraph::op::v0::ROIPooling>(input, reshape,
ngraph::Shape{(size_t)outHeight, (size_t)outWidth},
1.0f, "bilinear");
return Ptr<BackendNode>(new InfEngineNgraphNode(roiPooling));
}
#endif // HAVE_DNN_NGRAPH
#ifdef HAVE_CUDA
Ptr<BackendNode> initCUDA(
void *context_,
modules/dnn/src/layers/resize_layer.cpp
+4 -3
View File
@@ -41,7 +41,7 @@ public:
CV_Assert(params.has("zoom_factor_x") && params.has("zoom_factor_y"));
}
interpolation = params.get<String>("interpolation");
CV_Assert(interpolation == "nearest" || interpolation == "bilinear");
CV_Assert(interpolation == "nearest" || interpolation == "opencv_linear" || interpolation == "bilinear");
alignCorners = params.get<bool>("align_corners", false);
}
@@ -115,14 +115,15 @@ public:
Mat& inp = inputs[0];
Mat& out = outputs[0];
if (interpolation == "nearest")
if (interpolation == "nearest" || interpolation == "opencv_linear")
{
InterpolationFlags mode = interpolation == "nearest" ? INTER_NEAREST : INTER_LINEAR;
for (size_t n = 0; n < inputs[0].size[0]; ++n)
{
for (size_t ch = 0; ch < inputs[0].size[1]; ++ch)
{
resize(getPlane(inp, n, ch), getPlane(out, n, ch),
Size(outWidth, outHeight), 0, 0, INTER_NEAREST);
Size(outWidth, outHeight), 0, 0, mode);
}
}
}
modules/dnn/src/layers/scale_layer.cpp
+21 -16
View File
@@ -61,7 +61,8 @@ public:
return backendId == DNN_BACKEND_OPENCV ||
backendId == DNN_BACKEND_CUDA ||
backendId == DNN_BACKEND_HALIDE ||
((backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 || backendId == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH) && axis == 1);
(backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && axis == 1) ||
(backendId == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH && axis > 0);
}
void forward(InputArrayOfArrays inputs_arr, OutputArrayOfArrays outputs_arr, OutputArrayOfArrays internals_arr) CV_OVERRIDE
@@ -263,22 +264,26 @@ public:
auto ieInpNode = nodes[0].dynamicCast<InfEngineNgraphNode>()->node;
std::vector<size_t> shape(ieInpNode->get_shape().size(), 1);
shape[1] = numChannels;
auto weight = hasWeights ?
std::make_shared<ngraph::op::Constant>(ngraph::element::f32,
ngraph::Shape(shape), blobs[0].data) :
std::make_shared<ngraph::op::Constant>(ngraph::element::f32,
ngraph::Shape(shape), std::vector<float>(numChannels, 1).data());
int cAxis = clamp(axis, shape.size());
shape[cAxis] = numChannels;
auto bias = hasBias ?
std::make_shared<ngraph::op::Constant>(ngraph::element::f32,
ngraph::Shape(shape), blobs.back().data) :
std::make_shared<ngraph::op::Constant>(ngraph::element::f32,
ngraph::Shape(shape), std::vector<float>(numChannels, 0).data());
auto scale_node = std::make_shared<ngraph::op::v1::Multiply>(ieInpNode, weight, ngraph::op::AutoBroadcastType::NUMPY);
auto scale_shift = std::make_shared<ngraph::op::v1::Add>(scale_node, bias, ngraph::op::AutoBroadcastType::NUMPY);
return Ptr<BackendNode>(new InfEngineNgraphNode(scale_shift));
auto node = ieInpNode;
if (hasWeights)
{
auto weight = std::make_shared<ngraph::op::Constant>(ngraph::element::f32,
ngraph::Shape(shape), blobs[0].data);
node = std::make_shared<ngraph::op::v1::Multiply>(node, weight, ngraph::op::AutoBroadcastType::NUMPY);
}
if (hasBias || !hasWeights)
{
auto bias = hasBias ?
std::make_shared<ngraph::op::Constant>(ngraph::element::f32,
ngraph::Shape(shape), blobs.back().data) :
std::make_shared<ngraph::op::Constant>(ngraph::element::f32,
ngraph::Shape(shape), std::vector<float>(numChannels, 0).data());
node = std::make_shared<ngraph::op::v1::Add>(node, bias, ngraph::op::AutoBroadcastType::NUMPY);
}
return Ptr<BackendNode>(new InfEngineNgraphNode(node));
}
#endif // HAVE_DNN_NGRAPH
modules/dnn/src/onnx/onnx_importer.cpp
+24 -8
View File
@@ -485,16 +485,23 @@ void ONNXImporter::populateNet(Net dstNet)
}
else if (layer_type == "Split")
{
DictValue splits = layerParams.get("split");
const int numSplits = splits.size();
CV_Assert(numSplits > 1);
std::vector<int> slicePoints(numSplits - 1, splits.get<int>(0));
for (int i = 1; i < splits.size() - 1; ++i)
if (layerParams.has("split"))
{
slicePoints[i] = slicePoints[i - 1] + splits.get<int>(i - 1);
DictValue splits = layerParams.get("split");
const int numSplits = splits.size();
CV_Assert(numSplits > 1);
std::vector<int> slicePoints(numSplits - 1, splits.get<int>(0));
for (int i = 1; i < splits.size() - 1; ++i)
{
slicePoints[i] = slicePoints[i - 1] + splits.get<int>(i - 1);
}
layerParams.set("slice_point", DictValue::arrayInt(&slicePoints[0], slicePoints.size()));
}
else
{
layerParams.set("num_split", node_proto.output_size());
}
layerParams.set("slice_point", DictValue::arrayInt(&slicePoints[0], slicePoints.size()));
layerParams.type = "Slice";
}
else if (layer_type == "Add" || layer_type == "Sum")
@@ -973,6 +980,15 @@ void ONNXImporter::populateNet(Net dstNet)
replaceLayerParam(layerParams, "width_scale", "zoom_factor_x");
}
replaceLayerParam(layerParams, "mode", "interpolation");
if (layerParams.get<String>("interpolation") == "linear" && framework_name == "pytorch") {
layerParams.type = "Resize";
Mat scales = getBlob(node_proto, constBlobs, 1);
CV_Assert(scales.total() == 4);
layerParams.set("interpolation", "opencv_linear");
layerParams.set("zoom_factor_y", scales.at<float>(2));
layerParams.set("zoom_factor_x", scales.at<float>(3));
}
}
else if (layer_type == "LogSoftmax")
{
modules/dnn/test/test_ie_models.cpp
+20 -26
View File
@@ -73,28 +73,7 @@ struct OpenVINOModelTestCaseInfo
static const std::map<std::string, OpenVINOModelTestCaseInfo>& getOpenVINOTestModels()
{
static std::map<std::string, OpenVINOModelTestCaseInfo> g_models {
#if INF_ENGINE_RELEASE <= 2018050000
{ "age-gender-recognition-retail-0013", {
"deployment_tools/intel_models/age-gender-recognition-retail-0013/FP32/age-gender-recognition-retail-0013",
"deployment_tools/intel_models/age-gender-recognition-retail-0013/FP16/age-gender-recognition-retail-0013"
}},
{ "face-person-detection-retail-0002", {
"deployment_tools/intel_models/face-person-detection-retail-0002/FP32/face-person-detection-retail-0002",
"deployment_tools/intel_models/face-person-detection-retail-0002/FP16/face-person-detection-retail-0002"
}},
{ "head-pose-estimation-adas-0001", {
"deployment_tools/intel_models/head-pose-estimation-adas-0001/FP32/head-pose-estimation-adas-0001",
"deployment_tools/intel_models/head-pose-estimation-adas-0001/FP16/head-pose-estimation-adas-0001"
}},
{ "person-detection-retail-0002", {
"deployment_tools/intel_models/person-detection-retail-0002/FP32/person-detection-retail-0002",
"deployment_tools/intel_models/person-detection-retail-0002/FP16/person-detection-retail-0002"
}},
{ "vehicle-detection-adas-0002", {
"deployment_tools/intel_models/vehicle-detection-adas-0002/FP32/vehicle-detection-adas-0002",
"deployment_tools/intel_models/vehicle-detection-adas-0002/FP16/vehicle-detection-adas-0002"
}}
#else
#if INF_ENGINE_RELEASE >= 2018050000
// layout is defined by open_model_zoo/model_downloader
// Downloaded using these parameters for Open Model Zoo downloader (2019R1):
// ./downloader.py -o ${OPENCV_DNN_TEST_DATA_PATH}/omz_intel_models --cache_dir ${OPENCV_DNN_TEST_DATA_PATH}/.omz_cache/ \
@@ -118,7 +97,16 @@ static const std::map<std::string, OpenVINOModelTestCaseInfo>& getOpenVINOTestMo
{ "vehicle-detection-adas-0002", {
"Transportation/object_detection/vehicle/mobilenet-reduced-ssd/dldt/vehicle-detection-adas-0002",
"Transportation/object_detection/vehicle/mobilenet-reduced-ssd/dldt/vehicle-detection-adas-0002-fp16"
}}
}},
#endif
#if INF_ENGINE_RELEASE >= 2020010000
// Downloaded using these parameters for Open Model Zoo downloader (2020.1):
// ./downloader.py -o ${OPENCV_DNN_TEST_DATA_PATH}/omz_intel_models --cache_dir ${OPENCV_DNN_TEST_DATA_PATH}/.omz_cache/ \
// --name person-detection-retail-0013
{ "person-detection-retail-0013", { // IRv10
"intel/person-detection-retail-0013/FP32/person-detection-retail-0013",
"intel/person-detection-retail-0013/FP16/person-detection-retail-0013"
}},
#endif
};
@@ -305,8 +293,8 @@ TEST_P(DNNTestOpenVINO, models)
OpenVINOModelTestCaseInfo modelInfo = it->second;
std::string modelPath = isFP16 ? modelInfo.modelPathFP16 : modelInfo.modelPathFP32;
std::string xmlPath = findDataFile(modelPath + ".xml");
std::string binPath = findDataFile(modelPath + ".bin");
std::string xmlPath = findDataFile(modelPath + ".xml", false);
std::string binPath = findDataFile(modelPath + ".bin", false);
std::map<std::string, cv::Mat> inputsMap;
std::map<std::string, cv::Mat> ieOutputsMap, cvOutputsMap;
@@ -316,13 +304,19 @@ TEST_P(DNNTestOpenVINO, models)
runIE(targetId, xmlPath, binPath, inputsMap, ieOutputsMap);
runCV(backendId, targetId, xmlPath, binPath, inputsMap, cvOutputsMap);
double eps = 0;
#if INF_ENGINE_VER_MAJOR_GE(2020010000)
if (targetId == DNN_TARGET_CPU && checkHardwareSupport(CV_CPU_AVX_512F))
eps = 1e-5;
#endif
EXPECT_EQ(ieOutputsMap.size(), cvOutputsMap.size());
for (auto& srcIt : ieOutputsMap)
{
auto dstIt = cvOutputsMap.find(srcIt.first);
CV_Assert(dstIt != cvOutputsMap.end());
double normInf = cvtest::norm(srcIt.second, dstIt->second, cv::NORM_INF);
EXPECT_EQ(normInf, 0);
EXPECT_LE(normInf, eps) << "output=" << srcIt.first;
}
}
modules/dnn/test/test_onnx_importer.cpp
+15
View File
@@ -335,6 +335,9 @@ TEST_P(Test_ONNX_layers, Padding)
TEST_P(Test_ONNX_layers, Resize)
{
testONNXModels("resize_nearest");
if (backend == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019)
applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_NN_BUILDER);
testONNXModels("resize_bilinear");
}
TEST_P(Test_ONNX_layers, MultyInputs)
@@ -411,6 +414,18 @@ TEST_P(Test_ONNX_layers, ReduceL2)
testONNXModels("reduceL2");
}
TEST_P(Test_ONNX_layers, Split)
{
if (backend == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019)
applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_NN_BUILDER);
if (backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH)
applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_NGRAPH);
testONNXModels("split_1");
testONNXModels("split_2");
testONNXModels("split_3");
testONNXModels("split_4");
}
TEST_P(Test_ONNX_layers, Slice)
{
#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_LT(2019010000)
modules/dnn/test/test_tf_importer.cpp
+27 -4
View File
@@ -994,8 +994,16 @@ TEST(Test_TensorFlow, two_inputs)
normAssert(out, firstInput + secondInput);
}
TEST(Test_TensorFlow, Mask_RCNN)
TEST_P(Test_TensorFlow_nets, Mask_RCNN)
{
static const double kMaskThreshold = 0.5;
if (backend == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && target == DNN_TARGET_MYRIAD)
applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD, CV_TEST_TAG_DNN_SKIP_IE_NN_BUILDER);
if (target == DNN_TARGET_MYRIAD && getInferenceEngineVPUType() == CV_DNN_INFERENCE_ENGINE_VPU_TYPE_MYRIAD_X)
applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD_X);
applyTestTag(CV_TEST_TAG_MEMORY_1GB, CV_TEST_TAG_DEBUG_VERYLONG);
Mat img = imread(findDataFile("dnn/street.png"));
std::string proto = findDataFile("dnn/mask_rcnn_inception_v2_coco_2018_01_28.pbtxt");
@@ -1006,7 +1014,8 @@ TEST(Test_TensorFlow, Mask_RCNN)
Mat refMasks = blobFromNPY(path("mask_rcnn_inception_v2_coco_2018_01_28.detection_masks.npy"));
Mat blob = blobFromImage(img, 1.0f, Size(800, 800), Scalar(), true, false);
net.setPreferableBackend(DNN_BACKEND_OPENCV);
net.setPreferableBackend(backend);
net.setPreferableTarget(target);
net.setInput(blob);
@@ -1020,7 +1029,10 @@ TEST(Test_TensorFlow, Mask_RCNN)
Mat outDetections = outs[0];
Mat outMasks = outs[1];
normAssertDetections(refDetections, outDetections, "", /*threshold for zero confidence*/1e-5);
double scoreDiff = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 0.019 : 2e-5;
double iouDiff = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 0.018 : default_lInf;
normAssertDetections(refDetections, outDetections, "", /*threshold for zero confidence*/1e-5, scoreDiff, iouDiff);
// Output size of masks is NxCxHxW where
// N - number of detected boxes
@@ -1044,7 +1056,18 @@ TEST(Test_TensorFlow, Mask_RCNN)
outMasks(srcRanges).copyTo(masks(dstRanges));
}
cv::Range topRefMasks[] = {Range::all(), Range(0, numDetections), Range::all(), Range::all()};
normAssert(masks, refMasks(&topRefMasks[0]));
refMasks = refMasks(&topRefMasks[0]);
// make binary masks
cv::threshold(masks.reshape(1, 1), masks, kMaskThreshold, 1, THRESH_BINARY);
cv::threshold(refMasks.reshape(1, 1), refMasks, kMaskThreshold, 1, THRESH_BINARY);
double inter = cv::countNonZero(masks & refMasks);
double area = cv::countNonZero(masks | refMasks);
EXPECT_GE(inter / area, 0.99);
if (backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH)
expectNoFallbacks(net);
}
}